KVM: s390: Fix race condition in kvm_s390_handle_wait
The call to add_timer was issued before local_int.lock was taken and before
timer_due was set to 0. If the timer expires before the lock is being taken,
the timer function will set timer_due to 1 and exit before the vcpu falls
asleep. Depending on other external events, the vcpu might sleep forever.
This fix pulls setting timer_due to the beginning of the function before
add_timer, which ensures correct behavior.
Signed-off-by: Carsten Otte <cotte@de.ibm.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c
index fcd1ed8..84a7fed 100644
--- a/arch/s390/kvm/interrupt.c
+++ b/arch/s390/kvm/interrupt.c
@@ -339,6 +339,11 @@
if (kvm_cpu_has_interrupt(vcpu))
return 0;
+ __set_cpu_idle(vcpu);
+ spin_lock_bh(&vcpu->arch.local_int.lock);
+ vcpu->arch.local_int.timer_due = 0;
+ spin_unlock_bh(&vcpu->arch.local_int.lock);
+
if (psw_interrupts_disabled(vcpu)) {
VCPU_EVENT(vcpu, 3, "%s", "disabled wait");
__unset_cpu_idle(vcpu);
@@ -366,8 +371,6 @@
no_timer:
spin_lock_bh(&vcpu->arch.local_int.float_int->lock);
spin_lock_bh(&vcpu->arch.local_int.lock);
- __set_cpu_idle(vcpu);
- vcpu->arch.local_int.timer_due = 0;
add_wait_queue(&vcpu->arch.local_int.wq, &wait);
while (list_empty(&vcpu->arch.local_int.list) &&
list_empty(&vcpu->arch.local_int.float_int->list) &&